Home » Library » Modern Library » Alvin Plantinga Against Evil

Alvin Plantinga Against Evil

Objections to Draper’s Argument from Evil (2007)

Alvin Plantinga


          I. The Argument

          II. Challenges

          III. Is a Challenge a Problem?

I. The Argument

In his interesting and challenging paper, Paul Draper proposes to set out a “serious argument from evil against theism in which evolutionary theory plays a significant role.” He takes theism to be the proposition that the natural world has been created by an omniscient, omnipotent and morally perfect being; he takes naturalism to be the proposition that the natural world is closedsuch that nothing that is not part of the natural world has any effect on it. He then proposes to argue three things: (a) naturalism has smaller scope (less content) than theism, (b) naturalism is simpler than theism, and (c) naturalism has much greater predictive power than theism with respect to E, which states, roughly, that there is a great deal of suffering connected with the fact (as he sees it) that our living world has come to be by way of the processes endorsed by current evolutionary theory. He concludes that “naturalism is much more probable than theism, all else held equal. And that entails that, all else held equal, theism is very probably false.”

Say that a belief P is subject to a Draper Challenge (‘Challenge,’ for short) when there is a proposition Q incompatible with P that is simpler and has less scope than P, and a true proposition R that is much more probable with respect to Q than with respect to P. If a belief is subject to a Challenge, then, he says, all else being equal, it is very probably false; and of course what Draper argues is that theism is subject to a Challenge. This looks initially like a serious problem; what can the theist say for herself? I’ll argue that there is less, here, than meets the eye. That is because (as I’ll argue) very many, indeed most of the propositions P we believe are subject to Draper Challenges. If I am right, the disability with which Draper charges theism is widely shared, in fact shared by most of what we believe.

But first, just a word about scope and simplicity. “Roughly speaking,” says Draper, “scope is a measure of how much a hypothesis purports to tell us about the contingent world”; and if p has larger scope than q (tells us more about the contingent world than q), then p is less likely, all else being equal, than q. We might put this as follows: if p has larger scope than q, then p is true in fewer possible worlds than q.[1] Now Draper points out that according to theism, “all natural entities share a single ultimate supernatural (necessary) cause.” According to theism, that is, God, the ultimate cause of all natural entities, is a necessary beingone that exists in all possible worlds. This is certainly so according to most varieties of theism. But that means that the proposition there is such a person as God is true in all possible worlds, and hence has minimum content. Further, according to those same brands of theism, God has his central propertiesomniscience, omnipotence, moral perfectionessentially; he has them in every world in which he exists. Therefore (according to these versions of theism) the proposition there is an omniscient, omnipotent and morally perfect being is true in every possible world, has minimal scope, and has an intrinsic logical probability of 1. Still further, according to these same brands of theism, it isn’t possible that there be a contingent being that isn’t created by God (directly or indirectly); nothing can just pop uncaused into existence. But that means that in every possible world in which there are natural beings, in Draper’s sense, those natural beings are created by a supernatural being. It means further that naturalism is necessarily false. Naturalism, therefore, will be maximally improbable; theism, on the other hand, will be true in every world in which there are natural beings. According to these brands of theism, therefore, theism has vastly less scope and content than naturalism. But then Draper’s claim that naturalism has smaller scope than theism is true only if the most common varieties of theism are false. His argument, therefore, assumes from the start that the most common varieties of theism are false. But isn’t that dialectically deficient, something in the near neighborhood of begging the question? In any event, what he says doesn’t in any way constitute an argument against theism, so understood.

There are equally interesting things to say about simplicity as a mark of truth, but I don’t have the space to say them (there’s that stingy editor Draper mentions…). Let me just mention two of them, without going on to develop them. According to Aristotle, probability is Janus-faced: on the one hand, “a probability is that which happens for the most part,” and on the other a probability is a proposition that is approvable, worthy of belief.[2] The first is objective probability: logical, statistical, or some amalgam. The second is epistemic probability, the epistemic probability of a proposition (in a given context) being the rational or reasonable thing to believe (in that context).[3] Now simplicity, obviously, has something to do with epistemic probability. Draper also (apparently) thinks it is important for objective probability; the simple, he says, is a mark, not just of the right thing to believe, but of the truth itself. The reason for this is that “our reliance on inductive reasoning is clearly justified, and this is possible only if objective uniformity [simplicity–AP] either over time or at a single time, is intrinsically more probable than change or variety.”

Now for the two points about simplicity (as I say, I won’t have the time for more than bare assertion). First, it isn’t exactly simplicity that is at issue for epistemic probability; what counts is something more like naturalness, where the latter is to be explained in terms of what a properly functioning human being would think or believe in the relevant circumstance (WPF pp. 3-17). Draper proposes to gloss simplicity in terms of uniformity (“either over time or at a single time”). But what is uniformity? A diachronic situation is uniform if it doesn’t involve change, i.e., doesn’t involve something’s having a given property P at a time t, and at some later time failing to have P. But things always change in some respectse.g., with respect to how long they have been in existence: which respects are relevant? Well, being grue (gruesomeness?), says Draper (footnote 8), is not a relevant respect: a thing’s changing from green to blue involves an objective change, but a thing’s changing from grue to bleen doesn’t (in fact it involves an objective uniformity). But why say (or think) a thing like that? True: grue and bleen are ordinarily defined in terms of blue and green; but blue and green can also be defined in terms of grue and bleen. The difference between blue and green, on the one hand, and grue and bleen, on the other, I suggest, is a difference in epistemic naturalness: we human beings are naturally inclined to think in terms of blue and green but not grue and bleen; we naturally project the former but not the latter. Blue and green are epistemically natural for us, while grue and bleen are unnatural, repellent, repugnant, disgusting, and hard to employ. This is a feature of our cognitive architecture, but it isn’t cognitively inevitable. There could certainly be creatures who naturally thought in terms of grue and bleen, defining green and blue in terms of them. For them, what is epistemically probable (i.e., what a properly functioning creature of that sort would think in a given situation) might be quite different from what it is for us.

So let’s suppose epistemic probability depends in part on naturalness. Now there is a good deal of empirical research according to which belief in God, or at any rate supernatural beings, is natural for human beings.[4] Indeed, this is evident just from the nearly universal distribution of religion and religious beliefs across the world. But then if what is relevant here is not just simplicity but naturalness, theism would presumably be, so far forth, epistemically more probable than naturalism.

Second point: as we have seen, Draper believes that simplicity (uniformity), clearly a mark of epistemic probability, is also a mark of truth and hence a mark of objective probability. Simpler propositions are true in more possible worlds than complex propositions; hence they have greater objective probability. Why should we think this? Draper answers that “our reliance on inductive reasoning is clearly justified, and this is possible only if objective uniformity [i.e., simplicity–AP] either over time or at a single time, is intrinsically more probable than change or variety.” But this doesn’t seem entirely accurate. It could easily be that complex propositions were more probable (objectively) than simple ones, and that nonetheless our ways of thinking, which involve preferring simple to complex propositions, are justified, rational, warranted and reliable. That could be, for example, if God has created us in such a way that we prefer simple to complex propositions, and also created our world in such a way that (in our world) simplicity is a mark of truth (even though it isn’t in possible worlds generally).

II. Challenges

Draper argues that theism faces a Challenge: naturalism is both simpler and smaller in scope than theism, and also has much greater predictive power than theism with respect to that proposition E about the suffering and misery involved in that long evolutionary process. I’ve already argued that it is far from clear that theism has larger scope than naturalism: if God is a necessary being whose main properties are essential to him, as most theists who have thought about it have maintained, theism will have as small a scope as (will be as probable as) any proposition entailing that there are natural beings, while naturalism will be necessarily false. As for simplicity, I’ve argued that what is really at issue is epistemic naturalness; and here theism enjoys a clear advantage over naturalism.

But suppose we set these points aside. Further, suppose, just for purposes of argument, we concede Draper’s claim that naturalism is simpler and has less scope than theism, and also enjoys much greater predictive power than theism with respect to that proposition Ei.e., suppose we concede that theism suffers from a Challenge. Do I really have a serious problem when one of my beliefs is Challenged?

By way of looking into the matter, we should note first that many of the beliefs we actually hold are Challenged. We’re playing bridge: after the first hand is played, I believe I was dealt the 4, 7, 9, 10, queen and ace of spades, the 2, 5, 6, and queen of clubs, and the 3, 6, and 10 of diamonds, and that you were dealt the 3 of spades. You are not dealt any hearts. Now the probability that you were dealt no hearts, given that I’ve been dealt the cards I was dealt, is pretty low. So we have

P I was dealt the 4, 7, 9, 10, queen, and ace of spades, the 2, 5, 6, and queen of clubs and the 7, 8, and 9 of diamonds, and you were dealt the 3 of spades


R You were not dealt any hearts.

The probability of R on P is pretty low: if I wasn’t dealt any hearts, it’s very likely that you were dealt some. But there is another proposition in the neighborhood that is incompatible with P, simpler, has less scope, and has much greater predictive power with respect to R than P, namely

Q you were dealt only clubs and diamonds.

Although questions of scope are difficult, using what seem to be Draper’s criteria for scope, Q seems to have less scope than P; P is a rather detailed proposition asserting the existence of a very specific state of affairs; R is much less specific. Questions of simplicity are if anything more difficult; but if we equate simplicity with uniformity, as Draper seems to, then it would also seem that Q is simpler than P: P doesn’t seem to assert much by way of uniformity, and Q does assert a uniformity, namely that you were dealt only clubs and diamonds. And clearly Q has much greater predictive power with respect to R than P: P(R/P) is low and P(R/Q) = 1. But my learning that P is thus subject to a Challenge clearly doesn’t constitute an epistemic problem for me; the fact is I might have carefully examined and recorded the cards I was dealt, and also seen you play the 3 of spades. Then I would know P, despite its being subject to a Challenge.

Another example: some friends used to have an overweight Siamese cat named ‘Maynard’ who, oddly enough, liked green beans. So we have

P Maynard is an overweight Siamese cat who lives in South Bend, Indiana,

and for R we have

R Maynard likes green beans.

Here we can choose for Q

Q Maynard is a Frenchman.

Q is incompatible with P. Q also seems to be less specific than P, and hence is presumably of smaller scope. On most measures of simplicity, furthermore, Q would also be simpler than P. And P(R/Q) is much greater than P(R/P): the proportion of Frenchmen who like green beans is much higher than that of Siamese cats who do. But surely I don’t have an epistemic problem by virtue of believing P. The fact is I know P.

You might object that Q is impossible: it isn’t possible that something that is in fact a cat, should have been a human being, let alone a Frenchman. Well, perhaps so (although this isn’t just obvious: it depends upon the ontology of cats and human beings);[5] let’s add to Draper’s conditions that R must be contingent. Then consider instead

P Sam is a mean-tempered obese Frisian lifeguard who lives in Leeuwarden

R Sam is climbing El Cap


Q Sam is a climbing ranger in Yosemite.

Once again, Q is incompatible with P; Q also seems to have less scope than P (it is much more specific about Sam), and furthermore Q is plausibly thought to be simpler than P (in that P goes against a plausible uniformity, i.e., the uniformity that Frisian lifeguards are not obese). Finally Q has much greater (greater, one thinks, by several orders of magnitude) predictive power with respect to R than P: very few obese Frisian lifeguards climb El Cap.

My guess is that most of the propositions we believe are challenged.

III. Is a Challenge a Problem?

What sort of problem is a Challenge supposed to present? Draper says of theism, “all else held equal, theism is very probably false.” So the suggestion is that a belief subject to a Challenge is, all else held equal, very probably false. But how do we understand this “all else held equal”? Equal to what? What Draper means, I think, is this:

D If a belief P is subject to Challenge and the epistemic merits of the challenging belief Q are equal to those of P, then P is very likely false.

Now there is a problem with this ‘very likely false’: what is this probability conditioned on? What proposition X is it such that (under these conditions) P(P/X) is low? That’s not an easy question to answer, but let’s set it aside for the moment, and let’s assume that D is true. Let’s also suppose that the relevant epistemic merit is warrant, that quantity enough of which is sufficient to distinguish knowledge from mere true belief. And let’s suppose still further that the epistemic problem in question, the problem posed for someone S who holds a belief P that meets the conditions specified in the antecedent of D, is that S holds a belief for which he has a defeater. (More exactly, a “reflective rationality defeater,”[6] in that if he became aware that P meets those conditions, than he could no longer rationally believe P.) I’m sure Draper would point out that in the examples I cited in the last section, all else isn’t equal: the challenging belief doesn’t have nearly as much by way of warrant as the challenged belief. In those cases, therefore, a Challenge does not beget a defeater.

But then what about theism? Does the fact (as we are conceding for purposes of argument) that theism faces a Challenge mean that the theist has a defeater for her theism? I’d say it doesn’t. One kind of belief for which a Challenge begets a defeater is the kind of belief that get its warrant from its explaining or predicting some range of data; scientific hypotheses come to mind. If theism were a hypothesis of this sort, then perhaps it would be subject to defeat. But even here the defeater is only a prima facie defeater; it might be that a hypothesis H is subject to Challenge with respect to one proposition R, but survives that Challenge by explaining or predicting a whole range of other data better than alternatives. So even if theism were such a hypothesis, it might still be preserved from defeat by the fact (as I see it) that it explains many other thingsthat we have reliable faculties, that we can do philosophy, physics and evolutionary psychology, that there is such a thing as objective morality, that the universe is fine-tuned, that there are such things as propositions, properties, sets, and numbers, and the likemuch better than atheism.

But of course most theists don’t take it as a mere hypothesis designed to explain other things; most theists, I think, accept theism on the basis of experience of one sort or another. Indeed, if theism is true, perhaps theists know that it is true. In Warranted Christian Belief I argued that if theism is true, then very likely God would want human beings to be able to be aware of his presence and would have arranged for human beings, under certain conditions, to believe that theism is indeed true. If theism is true, then we human beings were created by God, who very likely intended that we have belief-producing processes that produce theistic belief. Under those conditions, these processes would presumably be functioning properly in producing such belief in us; furthermore they would be reliable processes, and would be successfully aimed at the production of true belief. If the belief in question is held with sufficient firmness, therefore, the belief constitutes knowledge.[7] Under those conditions, and for those people, therefore, theism wouldn’t be a hypothesis: it would be something they know, and know in the basic way. So the consequence is that if theism is true, then very likely there are some or many theists who know that it is; those theists, of course, wouldn’t have a defeater for theism in the fact (if it is a fact) that theism faces a Challenge. (D) would be irrelevant with respect to them, since they don’t satisfy the conditions of its antecedent. The same goes, presumably, for theists whose theistic beliefs have some warrant, but not warrant sufficient for knowledge; for them too presumably all else isn’t equal in that naturalism doesn’t have for them as much warrant as theism does.

If theism is true, therefore, the fact (if it is a fact) that it is subject to a Challenge does not constitute a problem for most theists. It is like my belief that Maynard is an overweight Siamese cat who lives in South Bend: that belief too is subject to a Challenge, but its being so subject doesn’t produce a defeater. In each case, the challenged belief has more warrant than the challenger. To produce a “serious argument from evil against theism in which evolutionary theory plays a significant role” along these lines, therefore, Draper would first have to show that theism is false; if it is true, then the fact that it is subject to Challenge won’t ordinarily produce a defeater. What really follows from the premises he has proposed, is that if theism is false, then theism is subject to a prima facie defeater by virtue of the Challenge Draper specifies. This is a conclusion in which the theist can happily acquiesce.

Continue the Debate


[1] Of course things are complicated by the fact that (as seems reasonable) there are at least uncountably many possible worlds. (Thus presumably for any real number n in the interval [70, 71], it is possible that you be n inches tall.) To mend matters, we can think of possible worlds as occupying a logical space; the intrinsic logical probability of a proposition p will be given by the proportion to the total logical space of the space occupied by the worlds in which p is true. (See John Bigelow, “Possible Worlds Foundations for Probability,” Journal of Philosophical Logic 5, no. 3 [1976]: 299-320.) The greater the content or scope of a proposition, the smaller the area occupied by the worlds in which p is true and the smaller the probability of p. A proposition with maximal scope (the book on a possible world, for example) will have minimal probability; a proposition with minimal scope (a necessary proposition, for example) will have maximal probability. Various problems arise here (e.g., is the appropriate measure on this space countably additive?); let us resolutely ignore them.

[2] Rhetoric I, 2 (1357 a 35, p. 2157) in J. Barnes, ed., The Complete Works of Aristotle (Princeton, NJ: Princeton University Press, 1984).

[3] See my Warrant and Proper Function (hereafter WPF) (New York: Oxford University Press, 1993), pp. 159 ff.

[4] See, e.g., Justin Barrett, “Exploring the Natural Foundations of Religion,” in Trends in Cognitive Sciences 4, no.1 (2000): 29-34; Barrett, Why Would Anyone Believe in God? (Lanham, MD: Altamira Press, 2004); Pascal Boyer, The Naturalness of Religious Ideas: A Cognitive Theory of Religion (Berkeley, CA: University of California Press, 1994); Boyer, Religion Explained: Evolutionary Origins of Religious Thought (New York: Basic Books, 2001); and Deborah Kelemen, “Are Children ‘Intuitive Theists’? Reasoning about Purpose and Design in Nature,” Psychological Science 15, no. 5 (2004): 295-301.

[5] See my The Nature of Necessity (Oxford: the Clarendon Press, 1974), “Could Socrates have been an Alligator?” pp. 65 ff.

[6] Roughly (very roughly) a rationality defeater of a belief B for a person S, is a belief B* S acquires such that one in S’s epistemic condition can’t rationally accept both B* and B. A reflective rationality defeater of a belief B for a person S is a proposition P such that if S were to reflect on her epistemic condition, she would come to believe P, thereby acquiring a rationality defeater for B. One who has a reflective rationality defeater for a belief B also has a warrant defeater for B.

[7] Assuming my account (WPF chaps. 1 and 2) of warrant. The same conclusion would follow, I think, on many other accounts of warrant.

Copyright ©2007 Alvin Plantinga. The electronic version is copyright ©2007 by Internet Infidels, Inc. with the written permission of Alvin Plantinga. All rights reserved.

all rights reserved